Skip to main content

Table 4 Comparative summary of the published studies

From: Would artificial neural networks implemented in clinical wards help nephrologists in predicting epoetin responsiveness?

Authors Number of patients/centres included Epoetin isoform Variables explored (inputs) Predictions (outputs)
Gabutti et al (present study) 432/29 Beta Sex; age; weight, presence or absence of a diabetes mellitus and/or a cardiomyopathy with EF<50%; haemoglobin; creatinine; BUN; pH; ionized calcium; albumin; CRP; ferritin; PTH; epoetin and iron dose; epoetin administration route sc vs. iv; Kt/V Epoetin dose and follow-up haemoglobin
Bellazzi [24] 10/1 n.s. Sex; age; haemoglobin; calcium; PTH; epoetin dose and others non specified Follow-up haemoglobin
Martin Guerriero et al [20] 110/1 Alpha and beta Age; weight; haemoglobin; ferritin; epoetin dose; isoform and number of administrations weekly; iron dose Epoetin dose
Gaweda et al [25] 209/1 n.s. Haematocrit; albumin; ferritin; iron saturation; PTH; epoetin and iron dose; Kt/V Follow-up haematocrit
Jacobs et al [26] 166/n.s. n.s. Haematocrit; albumin; ferritin; iron saturation; PTH; epoetin and iron dose; Kt/V Follow-up haematocrit
  1. n.s.: not stated
  2. Comparative table summarizing the characteristics of the studies already published predicting the epoetin dose and/or the follow-up haemoglobin/haematocrit with ANNs.